High flow nasal cannula for respiratory support in preterm infants
Abstract
Background
High flow nasal cannulae (HFNC) are small, thin, tapered binasal tubes that deliver oxygen or blended oxygen/air at gas flows of more than 1 L/min. HFNC are increasingly being used as a form of non‐invasive respiratory support for preterm infants.
Objectives
To compare the safety and efficacy of HFNC with other forms of non‐invasive respiratory support in preterm infants.
Search methods
We used the standard search strategy of the Cochrane Neonatal Review Group to search the Cochrane Central Register of Controlled Trials (CENTRAL 2016, Issue 1), MEDLINE via PubMed (1966 to 1 January 2016), EMBASE (1980 to 1 January 2016), and CINAHL (1982 to 1 January 2016). We also searched clinical trials databases, conference proceedings, and the reference lists of retrieved articles for randomised controlled trials and quasi‐randomised trials.
Selection criteria
Randomised or quasi‐randomised trials comparing HFNC with other non‐invasive forms of respiratory support in preterm infants immediately after birth or following extubation.
Data collection and analysis
The authors extracted and analysed data, and calculated risk ratio, risk difference and number needed to treat for an additional beneficial outcome.
Main results
We identified 15 studies for inclusion in the review. The studies differed in the interventions compared (nasal continuous positive airway pressure (CPAP), nasal intermittent positive pressure ventilation (NIPPV), non‐humidified HFNC, models for delivering HFNC), the gas flows used and the indications for respiratory support (primary support from soon after birth, post‐extubation support, weaning from CPAP support). When used as primary respiratory support after birth compared to CPAP (4 studies, 439 infants), there were no differences in the primary outcomes of death (typical risk ratio (RR) 0.36, 95% CI 0.01 to 8.73; 4 studies, 439 infants) or chronic lung disease (CLD) (typical RR 2.07, 95% CI 0.64 to 6.64; 4 studies, 439 infants). HFNC use resulted in longer duration of respiratory support, but there were no differences in other secondary outcomes. One study (75 infants) showed no differences between HFNC and NIPPV as primary support. Following extubation (total 6 studies, 934 infants), there were no differences between HFNC and CPAP in the primary outcomes of death (typical RR 0.77, 95% CI 0.43 to 1.36; 5 studies, 896 infants) or CLD (typical RR 0.96, 95% CI 0.78 to 1.18; 5 studies, 893 infants). There was no difference in the rate of treatment failure (typical RR 1.21, 95% CI 0.95 to 1.55; 5 studies, 786 infants) or reintubation (typical RR 0.91, 95% CI 0.68 to 1.20; 6 studies, 934 infants). Infants randomised to HFNC had reduced nasal trauma (typical RR 0.64, 95% CI 0.51 to 0.79; typical risk difference (RD) −0.14, 95% CI −0.20 to −0.08; 4 studies, 645 infants). There was a small reduction in the rate of pneumothorax (typical RR 0.35, 95% CI 0.11 to 1.06; typical RD −0.02, 95% CI −0.03 to −0.00; 5 studies 896 infants) in infants treated with HFNC. Subgroup analysis found no difference in the rate of the primary outcomes between HFNC and CPAP in preterm infants in different gestational age subgroups, though there were only small numbers of extremely preterm and late preterm infants. One trial (28 infants) found similar rates of reintubation for humidified and non‐humidified HFNC, and two other trials (100 infants) found no difference between different models of equipment used to deliver humidified HFNC. For infants weaning from non‐invasive respiratory support (CPAP), two studies (149 infants) found that preterm infants randomised to HFNC had a reduced duration of hospitalisation compared with infants who remained on CPAP.
Authors' conclusions
HFNC has similar rates of efficacy to other forms of non‐invasive respiratory support in preterm infants for preventing treatment failure, death and CLD. Most evidence is available for the use of HFNC as post‐extubation support. Following extubation, HFNC is associated with less nasal trauma, and may be associated with reduced pneumothorax compared with nasal CPAP. Further adequately powered randomised controlled trials should be undertaken in preterm infants comparing HFNC with other forms of primary non‐invasive support after birth and for weaning from non‐invasive support. Further evidence is also required for evaluating the safety and efficacy of HFNC in extremely preterm and mildly preterm subgroups, and for comparing different HFNC devices.
PICOs
Plain language summary
Nasal cannula for breathing support in premature babies
Review question: In preterm infants, is the use of high flow nasal cannulae (HFNC) as effective as other non‐invasive methods of respiratory support in preventing chronic lung injury and death?
Background: There are a variety of ways in which non‐invasive breathing support can be provided to preterm infants with irregular breathing (apnoea) or lung disease. These include supplemental oxygen given into the incubator, via a head‐box or via a nasal cannula; continuous positive airways pressure (CPAP) given via nasal prongs or mask; and nasal intermittent positive pressure ventilation (NIPPV) where, in addition to CPAP, inflations of a higher pressure are given intermittently. High flow nasal cannulae (HFNC) deliver oxygen or a mixture of oxygen and air via small, thin tubes that sit just inside the nostrils. HFNC have recently been introduced as another potential form of non‐invasive support.
Study characteristics: This review found 15 randomised studies that compared HFNC with other non‐invasive ways of supporting babies' breathing. The studies differed in the interventions that were compared, the gas flows used and the reasons for respiratory support.
Results: When HFNC was used as first‐line respiratory support after birth compared to CPAP (4 studies, 439 infants), there were no differences in the rates of death or chronic lung disease (CLD). HFNC use resulted in longer duration of respiratory support, but there were no differences in other outcomes. One study (75 infants) showed no differences between HFNC and NIPPV as breathing support after birth. When HFNC were used after a period of mechanical ventilation (total 6 studies, 934 infants), there were no differences between HFNC and CPAP in the rates of death or CLD. There was no difference in the rate of treatment failure or reintubation. Infants randomised to HFNC had less trauma to the infant's nose. There was a small reduction in the rate of pneumothorax in infants treated with HFNC. We found no difference between the effect of HFNC compared with CPAP in preterm infants in different gestational age subgroups, though there were only small numbers of extremely preterm and late preterm infants. One trial (28 infants) found similar rates of reintubation for humidified and non‐humidified HFNC, and two other trials (100 infants) found no difference between different models of equipment used to deliver humidified HFNC. For infants weaning from non‐invasive respiratory support (CPAP), two studies (149 infants) found that preterm infants randomised to HFNC had a reduced duration of hospitalisation compared with infants who remained on CPAP.
Conclusions: HFNC use has similar rates of efficacy to other forms of non‐invasive respiratory support in preterm infants for preventing treatment failure, death and CLD. Most evidence is available for the use of HFNC as post‐extubation support. Following extubation, use of HFNC is associated with less nasal trauma, and may be associated with reduced pneumothorax compared with nasal CPAP. Further adequately powered randomised controlled trials should be undertaken in preterm infants comparing HFNC with other forms of primary non‐invasive support after birth and for weaning from non‐invasive support. Further evidence is also required for evaluating the safety and efficacy of HFNC in extremely preterm and mildly preterm subgroups, and for comparing different HFNC devices.
Authors' conclusions
Background
There are a variety of ways in which respiratory support can be provided to preterm infants with apnoea or parenchymal lung disease non‐invasively, i.e. without an endotracheal tube . These include supplemental oxygen given into the incubator, via a head‐box or via a nasal cannula; continuous positive airways pressure (CPAP) given via nasal prongs or mask; and nasal intermittent positive pressure ventilation (NIPPV) where, in addition to CPAP, inflations of a higher pressure are given intermittently.
Nasal cannulae are small, thin, tapered tubes (usually less than 1 cm in length) that sit just inside each nostril without occluding them (Frey 2003). Oxygen delivered by 'low flow' nasal cannulae (LFNC) typically refers to the use of flow rates of less than or equal to 1 litre per minute (L/min). Usually the gas used is unblended (i.e. 100% oxygen), and is neither heated nor humidified. LFNC are commonly used in convalescing preterm infants, often with chronic lung disease (Walsh 2005). Use of LFNC does not appear to provide significant support to pulmonary function (apart from the provision of oxygen) (Hensey 2013; O'Donnell 2013).
In contrast, 'high‐flow' nasal cannulae (HFNC) deliver oxygen or blended oxygen and air at higher flow rates than LFNC. For the purposes of this review, HFNC delivery is defined as the use of gas flows greater than 1 L/min, although typically higher gas flows (e.g. 2 to 8 L/min) are used (Hough 2012; Manley 2012). Gas given via HFNC is routinely heated and humidified, as with CPAP. High gas flows in preterm infants may provide positive end‐expiratory pressure (PEEP) at similar levels to that commonly set with CPAP in clinical practice (Frey 2001; Sreenan 2001; Spence 2007; Wilkinson 2008; Lampland 2009). Washout of nasopharyngeal dead‐space has also been proposed as an important mechanism of action of HFNC (Dysart 2009; Frizzola 2011). In HFNC systems, circuit flow is adjusted according to clinical effect and, although a pressure relief valve is used in some circuits, the internal circuit pressure is not routinely measured. HFNC have been suggested as an alternative form of respiratory support for preterm infants with apnoea, respiratory distress syndrome or chronic lung disease. They appear to be easy to apply and maintain (Saslow 2006), and compared to CPAP they appear to be more comfortable for infants (Osman 2014), and are preferred by nurses (Roberts 2014) and parents (Klingenberg 2014).
Nasal CPAP is widely used in premature and term newborns and provides an effective, safe alternative to endotracheal intubation (Morley 2004). It has been shown to reduce extubation failure, treat apnoea and respiratory distress syndrome and, by minimising duration of mechanical ventilation, may reduce chronic lung disease (De Paoli 2003). The most effective and popular means of administering CPAP is by using short binasal prongs (Morley 2004). These prongs are designed to fit snugly into the infant's nostrils with minimal leakage. By contrast, nasal cannulae do not usually occlude the nostrils and have the potential for a large leak around them. Other methods of delivering CPAP to the nose that are in common use include single nasal prongs and nasal masks (De Paoli 2008). Oxygen administered by nasal CPAP is usually blended, humidified and heated. In contrast to HFNC, the pressure delivered by nasal CPAP circuits is directly measured and regulated.
Both CPAP and HFNC systems may have adverse effects in newborns. Binasal prongs used to deliver CPAP are associated with trauma to the nasal septum and distortion of the nares (Robertson 1996; Sreenan 2001). It has been thought that HFNC may cause less nasal injury (Saslow 2006), however the use of humidified, unheated HFNC has been associated with mucosal irritation, nasal obstruction or bleeding as well as a possible increase in the risk of nosocomial infection (Kopelman 2003a; Kopelman 2003b).
Concern has also been expressed about the possibility of lung overdistension and trauma from unmeasured and variable PEEP with HFNC (Finer 2005; Hegde 2013). One case associating HFNC with pneumocephalus, pneumo‐orbitis and scalp emphysema has been reported (Jasin 2008). Other possible risks associated with HFNC include gastric distension or perforation, as has been seen with CPAP (Garland 1985).
The purpose of this review is to compare HFNC with other methods of providing non‐invasive respiratory support in premature newborn infants.
Objectives
The objectives were as follows.
In preterm infants, to compare the efficacy and safety of HFNC with other non‐invasive methods of respiratory support including:
-
Ambient (head‐box or cot) oxygen;
-
Low flow nasal cannulae (LFNC);
-
Continuous positive airways pressure (CPAP), via nasal prongs or mask
-
Nasal intermittent positive pressure ventilation (NIPPV);
-
Alternative HFNC technique
Methods
Criteria for considering studies for this review
Types of studies
All randomised and quasi‐randomised studies (including crossover trials). Studies reported in abstract form were included in the 'Studies awaiting classification' category. Data from one unpublished study (published only in abstract form) were obtained from the authors to enable its inclusion in the review.
Types of participants
1. Preterm infants (< 37 weeks' gestational age) receiving respiratory support after birth, either prophylactically or for respiratory distress syndrome, without a prior period of intermittent positive pressure ventilation (IPPV).
2. Preterm infants (< 37 weeks' gestational age) receiving respiratory support following a period of intermittent positive pressure ventilation (IPPV).
Types of interventions
For the purposes of this review, we defined high flow nasal cannula oxygen as the delivery of oxygen or blended oxygen and air via nasal cannulae at gas flow rates greater than 1 L/min.
Alternative interventions included:
-
Head box oxygen;
-
Low flow nasal cannulae (gas flow rates less than or equal to 1 L/min);
-
Nasal CPAP;
-
NIPPV;
-
HFNC using an alternative technique (e.g. humidified versus non‐humidified, or different HFNC devices).
Types of outcome measures
Primary outcomes
-
Death (before hospital discharge) or chronic lung disease (as defined below);
-
Death;
-
Chronic lung disease. CLD was defined as a requirement for supplemental oxygen and/or respiratory support at 36 weeks' postmenstrual age (PMA) for infants born at less than 32 weeks' gestational age or at 28 days of age for infants born at 32 weeks' gestational age or later. Note: data from studies that reported an outcome of 'CLD' or 'bronchopulmonary dysplasia' ('BPD') without an accompanying definition were still included in this outcome.
Secondary outcomes
Treatment failure
-
Intubation (or re‐intubation) within 7 days of trial entry*. Note: studies that reported intubation (or re‐intubation) within a shorter period than 7 days (e.g. intubation < 72 hours) were included in the analysis of this outcome;
-
Treatment failure (as defined by the trial authors) within 7 days of trial entry*. Note: studies that reported treatment failure within a shorter period than 7 days (e.g. treatment failure < 72 hours) were included in the analysis of this outcome;
-
Intubation at any time point following trial entry*.
Respiratory support:
-
Duration of mechanical ventilation via an endotracheal tube (days, or post‐menstrual age (PMA) at end)*;
-
Duration of any form of respiratory support (mechanical ventilation, CPAP, high flow nasal cannulae, or oxygen) (days, or PMA at end);
-
Duration of hospitalisation (days, or PMA at end).
Complications:
-
Air leak syndromes (pneumothorax, pneumomediastinum, pneumopericardium or pulmonary interstitial emphysema (PIE)) reported either individually or as a composite outcome;
-
Nasal trauma (defined as erythema or erosion of the nasal mucosa, nares or septum). Note some studies reported this as a continuous outcome and were not able to be included in meta‐analysis;
-
Nosocomial sepsis (defined as positive blood or cerebrospinal fluid (CSF) cultures taken after five days of age). Note some studies used alternative definitions, or did not define sepsis. These were included in meta‐analysis;
-
Gastrointestinal perforation or severe necrotising enterocolitis (NEC) (stage II or more according to Bell's criteria (Bell 1978)). Note: some included studies only reported the incidence of NEC, and were included in the analysis of this outcome;
-
Weight gain prior to discharge from hospital;
-
Days to attain full feeds*.
Neurosensory outcomes:
-
Retinopathy of prematurity (ROP): any stage and stage 3 or greater;
-
Long‐term neurodevelopmental outcome (rates of cerebral palsy on physician assessment, developmental delay i.e. IQ 2 standard deviations less than the mean on validated assessment tools such as Bayley's Mental Developmental Index), blindness, hearing impairment requiring amplification.
Outcome measures that were not in the original review, that were modified or included after review of the available data, are marked with an asterisk (*).
Search methods for identification of studies
We used the criteria and standard methods of Cochrane and the Cochrane Neonatal Review Group (see the Cochrane Neonatal Group search strategy for specialized register).
We conducted a comprehensive search including: Cochrane Central Register of Controlled Trials (CENTRAL 2016, Issue 1) in The Cochrane Library; MEDLINE via PubMed (1996 to 1 January 2016); EMBASE (1980 to 1 January 2016); CINAHL (1982 to 1 January 2016) using the following search terms: (oxygen OR positive pressure) AND (nasal cannula* OR nasal prong), plus database‐specific limiters for RCTs and neonates (see Appendix 1 for the full search strategies for each database). No language restrictions were applied.
We searched clinical trials registries for ongoing or recently completed trials (clinicaltrials.gov; the World Health Organization’s International Trials Registry and Platform www.who.int/ictrp/en/; and the ISRCTN Registry). In addition, the published abstracts of the Society for Pediatric Research and the European Society for Paediatric Research were searched (2000 to 2014).
Data collection and analysis
The standard methods of the Cochrane Neonatal Review Group were employed.
Selection of studies
We included all randomised and quasi‐randomised controlled trials fulfilling the selection criteria described in the previous section. The authors reviewed the results of the search and separately selected the studies for inclusion. The review authors resolved any disagreement by discussion.
Data extraction and management
At least two review authors independently performed trial searches, assessments of methodology and extraction of data; and compared and resolved any differences found at each stage. For each trial, we collected information regarding blinding of randomisation, the intervention and outcome measurements as well as completeness of follow‐up. For crossover trials, data from the first period only were used.
Where any queries arose or where additional data were required, the study authors were contacted.
Assessment of risk of bias in included studies
We assessed the methodological quality of the studies using the following criteria: allocation concealment (blinding of randomisation), blinding of intervention, completeness of follow‐up, and blinding of outcome measurement or assessment. For each criterion, the assessment was one of the following: yes; no; can't tell. The review authors separately assessed each study and any disagreement was resolved by discussion. This information was added to the 'Characteristics of included studies' table.
The authors evaluated the following issues and entered them into the 'Risk of bias' table.
1) Sequence generation (checking for possible selection bias). Was the allocation sequence adequately generated?
For each included study, we categorised the method used to generate the allocation sequence as:
‐ adequate (any truly random process e.g. random number table; computerised random number generator);
‐ inadequate (any non‐random process e.g. odd or even date of birth; hospital or clinic record number);
‐ unclear.
(2) Allocation concealment (checking for possible selection bias). Was allocation adequately concealed?
For each included study, we categorised the method used to conceal the allocation sequence as:
‐ adequate (e.g. telephone or central randomisation; consecutively numbered sealed opaque envelopes);
‐ inadequate (open random allocation; unsealed or non‐opaque envelopes; alternation; date of birth);
‐ unclear.
(3) Blinding (checking for possible performance bias). Was knowledge of the allocated intervention adequately prevented during the study? At study entry? At the time of outcome assessment?
For each included study, we categorised the methods used to blind study participants and personnel from knowledge of which intervention a participant received. Blinding was assessed separately for different outcomes or classes of outcomes. We categorised the methods as:
‐ adequate, inadequate or unclear for participants;
‐ adequate, inadequate or unclear for personnel;
‐ adequate, inadequate or unclear for outcome assessors.
We classified objective outcomes (for example death, chronic lung disease) in the absence of blinding as unclear for performance bias. We classified subjective outcomes (for example nasal mucosal injury) in the absence of blinding as high risk for bias.
(4) Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, protocol deviations). Were incomplete outcome data adequately addressed?
For each included study and for each outcome, we described the completeness of data including attrition and exclusions from the analysis where possible. We noted whether attrition and exclusions were reported, the numbers included in the analysis at each stage (compared with the total number of randomised participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups or were related to outcomes. Where sufficient information was reported or supplied by the trial authors, we re‐included missing data in the analyses. We categorised the methods as:
‐ adequate (< 20% missing data);
‐ inadequate (≥ 20% missing data):
‐ unclear.
(5) Selective reporting bias. Are reports of the study free of any suggestion of selective outcome reporting?
For each included study, we described how we investigated the possibility of selective outcome reporting bias and what we found. We assessed the methods as:
‐ adequate (where it was clear that all of the study’s pre‐specified outcomes and all expected outcomes of interest to the review have been reported);
‐ inadequate (where not all the study’s pre‐specified outcomes were reported; one or more of the reported primary outcomes were not pre‐specified; outcomes of interest were reported incompletely and so could not be used; study failed to include results of a key outcome that would have been expected to have been reported);
‐ unclear.
(6) Other sources of bias. Was the study apparently free of other problems that could put it at a high risk of bias?
For each included study, we described any important concerns we had about other possible sources of bias (for example, whether there was a potential source of bias related to the specific study design or whether the trial was stopped early due to some data‐dependent process). We assessed whether each study was free of other problems that could put it at risk of bias, as:
‐ yes; no; or unclear.
If needed, we planned to explore the impact of the level of bias through undertaking sensitivity analyses.
Measures of treatment effect
We extracted categorical data (for example number dying or with chronic lung disease) for each intervention group, and calculated risk ratio (RR), risk difference (RD) and number needed to treat for an additional beneficial outcome (NNTB) or number needed to treat for an additional harmful outcome (NNTH) as appropriate. We obtained means and standard deviations for continuous data (for example number of days of respiratory support, or duration of oxygen dependency); and we calculated the 95% confidence interval (CI) for each measure of effect.
Assessment of heterogeneity
We estimated heterogeneity using the I² statistic.
Data synthesis
We applied the fixed‐effect model for meta‐analysis. We obtained means and standard deviations for continuous data (for example number of days of respiratory support, or duration of oxygen treatment) and performed analysis using the weighted mean difference (WMD). We calculated the 95% CI for each measure of effect.
Subgroup analysis and investigation of heterogeneity
Where there was sufficient data we performed subgroup analysis by gestational age (GA) at birth for the primary outcome (death or CLD) and its components, and for treatment failure:
-
GA > 32 weeks*
-
GA 28 to 32 weeks*
-
GA < 28 weeks*
Subgroups that were modified since the original protocol/review are marked with an asterisk (*).
Sensitivity analysis
We had planned to perform a sensitivity analysis for quality of methods used.
Results
Description of studies
See Characteristics of included studies, Characteristics of excluded studies
We identified 15 studies for inclusion. The studies by Campbell 2006 (40 infants), Woodhead 2006 (30 infants), Miller 2010 (40 infants), Abdel Hady 2011 (60 infants), Collins 2013 (132 infants), Manley 2013 (303 infants), Yoder 2013 (351 infants), Mostafa‐Gharehbaghi 2014 (85 infants), Sadeghnia 2014 (60 infants), Badiee 2015 (89 infants), and Kugelman 2015 (76 infants) were available as full journal publications. The study by Nair 2005 (67 infants) was published as an abstract; additional unpublished data were provided by the authors enabling its inclusion in this review. Iranpour 2011 (70 infants), Ciuffini 2014 (177 infants), and Liu 2014 (150 infants) were published in English in abstract form and in full text in Persian, Italian and Chinese respectively. The authors of Iranpour 2011, Collins 2013, Manley 2013, Yoder 2013, Mostafa‐Gharehbaghi 2014, Liu 2014, and Kugelman 2015 kindly provided additional data. Several randomised controlled trials of HFNC versus other means of non‐invasive support are currently in progress, have been completed but are not yet published, or are awaiting further assessment (NCT01939067; ACTRN12615000077561; IRCT2014012716376N1; Lawrence 2012; Chen 2015; Febre 2015; Tang 2015; NCT02055339; ACTRN12610000677000; ISRCTN66716753; ACTRN12613000303741; JPRN‐UMIN000013906; NCT01270581).
1. HFNC versus CPAP for primary respiratory support after birth
Nair 2005 was a single‐centre study that enrolled 67 preterm infants of 27 to 34 weeks' gestational age (GA) with respiratory distress in the first six hours after birth. Infants in this study had a mean GA of 32 weeks and birth weight of 1700 grams and were randomised to HFNC (mean flow rate 5 to 6 L/min) or CPAP (5 to 6 cmH2O). The primary outcome was respiratory failure requiring intubation, based on prespecified criteria.
Yoder 2013 was a multi‐centre study that enrolled 432 term and preterm infants of more than 28 weeks' GA who were planned to receive non‐invasive respiratory support either as primary support after birth or post‐extubation. Of these, 351 infants were preterm, with 125 preterm infants in the primary support arm, and 226 in the post‐extubation arm. Infants were randomised to HFNC (3 to 5 L/min) or nasal CPAP (5 to 6 cmH2O). The primary outcome was need for intubation within 72 hours of commencing the allocated treatment, based on prespecified criteria.
Iranpour 2011 was a single‐centre study, published in Persian, that enrolled 70 preterm infants of 30 to 35 weeks' gestation at 24 hours of age who had ongoing features of respiratory distress and oxygen requirement. Infants were randomised to HFNC (gas flow 1.5 to 3 L/min based on Sreenan 2001) or to continuing nasal CPAP (6 cmH2O). Infants who met prespecified criteria (before or after randomisation) received surfactant via an INSURE (Intubation, Surfactant administration, Extubation) technique.
Ciuffini 2014 was a report of interim results from a single‐centre study, published in Italian, that enrolled 177 of a planned 316 preterm infants, 29 to 36 weeks' GA, with mild to moderate respiratory distress after birth. Infants in the study had mean GA of 33 weeks and birth weight of 1900 grams. Infants were randomised to receive HFNC (4 to 6 L/min) or nasal CPAP (4 to 6 cmH2O). The primary outcome was the need for intubation within 72 hours of life, based on prespecified criteria.
2. HFNC versus NIPPV for primary respiratory support after birth
Kugelman 2015 was a single‐centre study that enrolled 76 preterm infants of less than 35 weeks' GA, with birth weight exceeding 1000 grams, who required primary non‐invasive respiratory support. Infants in the study had mean GA of 33 weeks and birth weight of 1800 grams. Infants were treated with either HFNC (starting gas flow 1 L/min, increased up to 5 L/min as required) or synchronised NIPPV (positive inflation pressure 14 to 22 cmH2O, positive end‐expiratory pressure 6 cmH2O, rate 12 to 30 inflations per minute). The primary outcome was treatment failure according to prespecified criteria.
3. HFNC versus CPAP to prevent extubation failure
Campbell 2006 was a single‐centre study that enrolled 40 intubated preterm infants (birth weight ≤ 1250 grams). Infants in this study had a mean GA of 27 weeks and birth weight of 1000 grams. Infants were randomised to humidified, unheated HFNC (mean gas flow 1.6 L/min) or variable flow CPAP (5 to 6 cmH2O) after extubation. The primary outcome was need for reintubation, based on prespecified criteria.
Collins 2013 was a single‐centre study that enrolled 132 intubated very preterm infants (< 32 weeks gestation at birth). Infants in the study had mean GA of 28 weeks and birth weight of 1100 grams. Infants were randomised to receive either HFNC (8 L/min) or nasal CPAP (8 cmH2O) after extubation. The primary outcome was extubation failure in the first seven days after extubation, based on prespecified criteria.
Manley 2013 was a multi‐centre, non‐inferiority study that enrolled 303 intubated very preterm infants (< 32 weeks' gestation at birth). Infants in the study had mean GA of 27 weeks and birth weight of 1000 grams. Infants were randomised to receive either HFNC (5 to 6 L/min) or CPAP (7 cmH2O) after extubation. The primary outcome was treatment failure within seven days of randomisation, based on prespecified criteria.
Yoder 2013 (see above) included 226 preterm infants enrolled post‐extubation.
Liu 2014 was a multi‐centre study, published in Chinese, that enrolled a total of 155 infants (< 7 days old), of which 150 were preterm. Infants in the study had a mean GA of 35.5 weeks, and birth weight of 2500 grams. Infants were randomised to either HFNC (gas flow 3 to 8 L/min depending on infant weight) or nasal CPAP (pressure as set pre‐extubation) after extubation. The primary outcomes were extubation failure (reintubation within seven days), BPD or death in hospital.
Mostafa‐Gharehbaghi 2014 was a single‐centre study that enrolled 123 preterm infants with GA of 30 to 34 weeks and birth weight of 1250 to 2000 grams. Infants in the study had mean GA of 32 weeks and birth weight of 1900 grams. Infants were initially stabilised with nasal CPAP and treated with intubation and surfactant in the NICU (INSURE technique). Infants were extubated after INSURE to either HFNC (6 L/min) or nasal CPAP (5 to 6 cmH2O). The primary outcome was re‐intubation within three days of surfactant administration, according to prespecified criteria.
4. Humidified HFNC versus non‐humidified HFNC to prevent extubation failure
Woodhead 2006 was a single‐centre study that enrolled 30 preterm infants. Infants in the study had a mean GA of 32 weeks and birth weight of 1700 grams. Infants were randomised to humidified HFNC (VapothermTM) (mean gas flow 3.1 L/min) or non‐humidified HFNC (mean gas flow 1.8 L/min) following extubation. This was a randomised crossover trial; results from only the first study period were used for analysis. The primary outcome was failure of extubation (defined either by the need for reintubation or a switch to the alternative modality of HFNC).
5. Alternative HFNC models to prevent extubation failure
Miller 2010 was a single‐centre pilot study that enrolled 40 preterm infants of 26 to 29 weeks' GA who had been intubated in the first 72 hours of life. The infants in the study had mean GA of 28 weeks and birth weight of 1100 grams. Infants were randomised to one of two different brands of equipment (Fischer and PaykelTM versus VapothermTM) for delivery of humidified HFNC at 6 L/min. The primary outcome was the need for reintubation within 72 hours of extubation, based on prespecified criteria.
Sadeghnia 2014 was a single centre study that enrolled 60 preterm infants (1000 to 1500 grams) who had previously received surfactant, and were stable on CPAP 4 cmH2O with supplemental oxygen requirement of less than 30%, but required supplemental oxygen when CPAP discontinued. Infants were randomised to HFNC at a gas flow based on Sreenan 2001 using two different humidifiers (MR850 vs PMH7000). The primary outcome (specified at study registration) was humidity of gas delivered.
6. HFNC for weaning from CPAP
Abdel Hady 2011 was a single‐centre study that enrolled preterm infants whose GA was 28 weeks and above, and who were stable on low levels of non‐invasive respiratory support (CPAP 5 cmH2O and supplemental oxygen ≤ 30%). Infants in the study had mean GA of 31 weeks and birth weight of 1600 grams. Infants were randomised to HFNC (2 L/min) or to remain on CPAP until no longer requiring supplemental oxygen. The primary outcome was the duration of supplemental oxygen and respiratory support.
Badiee 2015 was also a single centre study. It enrolled infants of 28 to 36 weeks' GA who were stable on CPAP 5cmH2O and less than 30% supplemental oxygen. Infants had a mean GA at birth of 31 weeks. They were randomised to HFNC (2 L/min) or to remain on CPAP. The primary outcome was the duration of supplemental oxygen.
Risk of bias in included studies
Blinding of treatment allocation was not attempted in any of the studies. There were preset criteria for treatment failure/intubation in all of the studies except Woodhead 2006. Treatment with the alternate intervention was not permitted in the first 72 hours in Yoder 2013 or Liu 2014; it was permitted in Abdel Hady 2011, Collins 2013, Manley 2013, Mostafa‐Gharehbaghi 2014, and Kugelman 2015.
Abdel Hady 2011, Collins 2013, Manley 2013, and Kugelman 2015 separately reported the incidence of treatment failure as well as intubation. In these studies, a number of infants in the HFNC group meeting treatment failure criteria were treated with 'rescue' CPAP or NIPPV and not subsequently intubated.
In some of the studies alterations to flow rates or the level of non‐invasive support were left to the discretion of treating clinicians (Woodhead 2006; Miller 2010; Collins 2013; Manley 2013; Ciuffini 2014; Kugelman 2015). This may have contributed to a difference in HFNC gas flows between the two arms of the study in Woodhead 2006.
Frequency of blood gas analysis and recording of apnoea frequency and severity were potentially open to bias. Lack of blinding was a potential source of bias for subjective outcomes such as the presence of nasal mucosal injury or abdominal distension. Only Woodhead 2006 reported blinded assessment of the nasal mucosa.
Secondary outcomes were retrieved from medical records in all studies and were potentially open to bias. One patient in the study by Miller 2010 was excluded from the analysis after developing sepsis and dying during the study, though this patient should have been included as requiring reintubation.
Allocation concealment was not clear in the studies by Woodhead 2006, Miller 2010, and Sadeghnia 2014.
Trial registration was not evident (or occurred after trial completion) for Nair 2005, Campbell 2006, Woodhead 2006, Miller 2010, Abdel Hady 2011, Iranpour 2011, Collins 2013, Ciuffini 2014, Liu 2014, Sadeghnia 2014 and Badiee 2015, raising the potential for selective reporting of outcomes. Ciuffini 2014 is a report of preliminary data from their trial prior to achieving the planned sample size.
Effects of interventions
Comparison 1. HFNC versus CPAP for primary respiratory support after birth
Four studies were available for this comparison (total 439 infants) (Nair 2005; Iranpour 2011; Yoder 2013; Ciuffini 2014). Rates of treatment failure (and need for intubation) within seven days of trial entry were similar between HFNC and CPAP (Figure 1). There were no differences in the rates of death (typical risk ratio (RR) 0.36, 95% confidence interval (CI) 0.01 to 8.73; 4 studies, 439 infants) or chronic lung disease (typical RR 2.07, 95% CI 0.64 to 6.64; 4 studies, 439 infants). The use of HFNC as primary support resulted in a longer duration of receiving respiratory support in one study (Yoder 2013). Other secondary outcomes (including nasal trauma, durations of supplemental oxygen and hospitalisation, pneumothorax, and sepsis) were similar between groups.
Forest plot of comparison: 1 HFNC versus CPAP soon after birth for treatment or prophylaxis of RDS, outcome: 1.4 Treatment failure within 7 days of trial entry.
Data on primary outcomes for gestational age subgroups were not available in Nair 2005, Iranpour 2011, or Ciuffini 2014. There were no differences in the primary outcomes or in treatment failure within GA subgroups from one study (Yoder 2013); however, there were only very small numbers of infants included in these subgroups and no extremely preterm infants (< 28 weeks' GA).
Comparison 2. HFNC versus NIPPV for primary respiratory support after birth
One study was available for this comparison (total 76 infants) (Kugelman 2015). There was no difference between HFNC and NIPPV in rates of treatment failure, death or CLD. Infants randomised to HFNC spent a longer period of time receiving non‐invasive respiratory support (median 4 days vs median 2 days, P < 0.01).
Comparison 3. HFNC versus CPAP to prevent extubation failure
Six studies were available for this comparison (total 934 infants) (Campbell 2006; Collins 2013; Manley 2013; Yoder 2013; Liu 2014; Mostafa‐Gharehbaghi 2014). Following extubation, there were no differences between HFNC and CPAP in the primary outcomes of death (typical RR 0.77, 95% CI 0.43 to 1.36; 5 studies, 896 infants) (Figure 2); or CLD (typical RR 0.96, 95% CI 0.78 to 1.18; 5 studies, 893 infants) (Figure 3). There was no difference in the rate of treatment failure (typical RR 1.21, 95% CI 0.95 to 1.55; 5 studies, 786 infants) (Figure 4); or reintubation (typical RR 0.91, 95% CI 0.68 to 1.20; 6 studies, 934 infants) (Figure 5). Infants randomised to HFNC had reduced nasal trauma (typical RR 0.64, 95% CI 0.51 to 0.79; typical risk difference (RD) −0.14, 95% CI −0.20 to −0.08; 4 studies, 645 infants) (Figure 6). There was a small reduction in the rate of pneumothorax (typical RR 0.35, 95% CI 0.11 to 1.06; typical RD −0.02, 95% CI −0.03 to −0.00; 5 studies, 896 infants) (Figure 7) in infants treated with HFNC. There was also an apparent small reduction in the rate of gastrointestinal perforation or severe NEC (typical RR 0.52, 95% CI 0.24 to 1.11; typical RD −0.02, 95% CI −0.05 to −0.00; 5 studies, 840 infants), though this did not reach statistical significance. There was no significant difference in the incidence of intraventricular haemorrhage, sepsis or ROP between groups.
Forest plot of comparison: 3 HFNC versus CPAP to prevent extubation failure, outcome: 3.3 Death.
Forest plot of comparison: 3 HFNC versus CPAP to prevent extubation failure, outcome: 3.2 CLD.
Forest plot of comparison: 3 High Flow Nasal Cannula versus CPAP to prevent extubation failure, outcome: Treatment failure.
Forest plot of comparison: 3 HFNC versus CPAP to prevent extubation failure within 7 days, outcome: 3.5 Reintubation within 7 days of trial entry.
Forest plot of comparison: 3 HFNC versus CPAP to prevent extubation failure, outcome: Nasal trauma.
Forest plot of comparison: 3 HFNC versus CPAP to prevent extubation failure, outcome: Pneumothorax.
Data on gestational age subgroups were available for five studies (Collins 2013; Manley 2013; Yoder 2013; Liu 2014; Mostafa‐Gharehbaghi 2014). There was no difference between HFNC and CPAP in the rate of death, CLD, or treatment failure in different subgroups. However, in infants from 28 to 32 weeks' gestation (the GA subgroup with the most data available) HFNC was associated with a significantly reduced rate of re‐intubation (typical RR 0.51, 95% CI 0.27 to 0.97; typical RD −0.06, 95% CI −0.12 to −0.00; 5 studies, 382 infants) (Figure 5). In several of these studies 'rescue' treatment with CPAP/NIPPV was used for infants randomised to HFNC meeting treatment failure criteria. One small study that did not have subgroup data available found a higher rate of reintubation in infants randomised to HFNC (Campbell 2006).
Comparison 4. Humidified HFNC versus non‐humidified HFNC to prevent extubation failure
One study was available for this comparison (Woodhead 2006). There was no significant difference in need for intubation during the first 24 hours of the study (prior to crossover) (0/15 infants treated with humidified HFNC compared to 2/15 infants treated with non‐humidified HFNC). Nasal mucosal scores were not available for the first study period, however the authors noted more infants in the humidified HFNC group had nasal mucosa with a normal appearance.
Comparison 5. Alternative HFNC models to prevent extubation failure
One study compared different HFNC models (Miller 2010). There was no significant difference in the need for reintubation within 72 hours of extubation (3/17 infants treated with Fisher and Paykel, 3/22 infants treated with Vapotherm). There was one death in the Vapotherm group, and one infant in the Fisher and Paykel group developed necrotising enterocolitis. Rates of chronic lung disease were similar between the two groups. Other secondary outcomes were not different or not reported.
One study compared different humidification devices for delivery of HFNC (Sadeghnia 2014). There was no significant difference in the need for mechanical ventilation, rate of CLD, or other secondary outcomes.
Comparison 6. HFNC for weaning from CPAP
Two studies compared the use of HFNC versus continued CPAP for weaning preterm infants who were stable on low levels of CPAP (Abdel Hady 2011; Badiee 2015). In one of these studies (Abdel Hady 2011), infants randomised to HFNC had a longer total duration of oxygen therapy (median 14 days vs median 5 days P < 0.001) and a longer period of respiratory support (median 18 days vs median 10.5 days, P < 0.05). Infants in the HFNC group had a slightly longer duration of respiratory support prior to randomisation (median 8.5 days, IQR 7 to 14.25) compared with the CPAP group (median 5.5 days, IQR 3 to 13, P = 0.07). In the second study (Badiee 2015), infants randomised to HFNC had a shorter duration of oxygen therapy (21 hours vs 50 hours); however, infants in this group commenced weaning at an earlier gestational age (32.2 weeks vs 33.6 weeks). Four babies in the CPAP group required intubation in one study, while no infants randomised to HFNC weaning required intubation. There was no difference in weaning failure, nor in major morbidities (sepsis, IVH, BPD). There was a small overall reduction in length of hospitalisation in infants receiving HFNC (typical RD −3.3 days, 95% CI −6.6 to 0.0 days; 2 studies, 149 infants).
Discussion
This review identified 15 randomised trials including a total of 1725 premature infants that compared respiratory support with high flow nasal cannulae (HFNC) with other forms of non‐invasive respiratory support in preterm infants. There were no studies comparing HFNC with ambient oxygen or low flow nasal cannulae (LFNC). Subgroup analysis by GA was only possible for the comparison of HFNC with continuous positive airway pressure (CPAP) for preventing extubation failure.
The 15 studies varied in study quality. None of the studies were blinded and bias may have occurred, particularly where there were no established criteria for treatment failure/reintubation, or where rescue treatment with other forms of respiratory support was permitted. We did not perform a sensitivity analysis for study quality.
HFNC for primary respiratory support after birth
For preterm infants needing primary respiratory support after birth, there were no differences in the rates of primary or secondary outcomes between HFNC and CPAP, or HFNC and nasal intermittent positive pressure ventilation (NIPPV). Four studies compared HFNC with CPAP, while only one study compared HFNC with NIPPV. Studies varied in the HFNC gas flows used and in the degree of prematurity of infants. Subgroup meta‐analysis was not possible, as data for GA subgroups were available for only one study (Yoder 2013)
HFNC for respiratory support after extubation
Eight studies evaluated HFNC as respiratory support post‐extubation. Overall, there was no difference in the rates of death or CLD in 934 preterm infants treated with HFNC or CPAP. There were no differences in the rates of treatment failure or reintubation. Infants treated with HFNC had a small reduction in the rate of pneumothorax (NNTB 50), and there was an apparent (though not significant) reduction in the rate of necrotising enterocolitis (NNTB 50). Subgroup analysis revealed a small reduction in the rate of reintubation in infants of 28 to 32 weeks' gestation treated with HFNC (NNTB 17). However, subgroup data were not available for all studies, and there were relatively few extremely preterm infants (< 28 weeks' GA) included in the studies we identified.
HFNC for weaning from CPAP
Two small studies assessed the use of HFNC for stable preterm infants weaning off respiratory support. Those studies found a small reduction in length of hospitalisation, but no difference in weaning failure or major morbidities..
Duration of support
Three studies included in this review reported a longer duration of weaning from respiratory support in the HFNC group. Abdel Hady 2011 found that infants randomised to HFNC (compared with those remaining on CPAP) received a longer total duration of respiratory support and a longer period of oxygen. Infants randomised to HFNC had a longer period of respiratory support prior to enrolment than those infants randomised to CPAP, which may have contributed to the difference. Kugelman 2015 found longer median duration of respiratory support compared with NIPPV. Yoder 2013 identified a longer duration of respiratory support compared with CPAP for preterm infants receiving HFNC as primary support (but not post‐extubation).
The significance of this finding is unclear. Other studies found no difference in the duration of respiratory support between HFNC and CPAP (Manley 2013; Collins 2013). Badiee 2015 found a lower duration of oxygen in infants weaned from CPAP using HFNC (2 L/min). Meta‐analysis was not possible, because of the different interventions compared and different methods for quantifying and reporting duration of support. It is possible that the lower flow rates of HFNC used in Abdel Hady 2011 (2 L/min) and Kugelman 2015 (starting flow rate 1 L/min) contributed to slower weaning.
Nasal trauma
Nasal trauma was less common in HFNC‐treated infants than infants treated with alternative means of respiratory support in seven of the studies included in this review (Nair 2005; Woodhead 2006; Iranpour 2011; Collins 2013; Manley 2013; Yoder 2013; Mostafa‐Gharehbaghi 2014); but no difference was seen in two ( Liu 2014; Kugelman 2015). Studies varied widely in the tools used to assess the severity of nasal injury. Only one study attempted to blind assessment of nasal mucosa (Woodhead 2006).
Different forms of HFNC
There was no evidence from one small study of benefit from humidification of HFNC (Woodhead 2006). During the second half of the crossover trial infants treated with non‐humidified HFNC had a higher rate of being switched to humidified HFNC because of perceived treatment failure. This may have related to the higher flow rates used in infants receiving humidified HFNC. There were higher (more abnormal) scores for nasal mucosal injury in infants treated with non‐humidified HFNC.
There was no difference in effectiveness between two different models of equipment used to deliver HFNC in two other small studies (Miller 2010; Sadeghnia 2014). None of the included studies examined the effect of different flow rates or cannula sizes.
Neurodevelopmental outcomes
No included studies reported long‐term neurodevelopmental outcomes.
Forest plot of comparison: 1 HFNC versus CPAP soon after birth for treatment or prophylaxis of RDS, outcome: 1.4 Treatment failure within 7 days of trial entry.
Forest plot of comparison: 3 HFNC versus CPAP to prevent extubation failure, outcome: 3.3 Death.
Forest plot of comparison: 3 HFNC versus CPAP to prevent extubation failure, outcome: 3.2 CLD.
Forest plot of comparison: 3 High Flow Nasal Cannula versus CPAP to prevent extubation failure, outcome: Treatment failure.
Forest plot of comparison: 3 HFNC versus CPAP to prevent extubation failure within 7 days, outcome: 3.5 Reintubation within 7 days of trial entry.
Forest plot of comparison: 3 HFNC versus CPAP to prevent extubation failure, outcome: Nasal trauma.
Forest plot of comparison: 3 HFNC versus CPAP to prevent extubation failure, outcome: Pneumothorax.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 1 Death or CLD.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 2 CLD.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 3 Death.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 4 Treatment failure within 7 days of trial entry.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 5 Intubation within 7 days of trial entry.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 6 Intubation at any time point after trial entry.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 7 Duration of any respiratory support (days).
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 8 Duration of supplemental oxygen (days).
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 9 Duration of hospitalisation (days).
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 10 Pneumothorax.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 11 Nosocomial sepsis.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 12 Gastrointestinal perforation or severe NEC.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 13 Days to full feeds.
Comparison 1 HFNC versus CPAP for primary respiratory support after birth, Outcome 14 Nasal trauma.
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 1 Death or CLD.
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 2 CLD.
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 3 Death.
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 4 Treatment failure within 7 days of trial entry.
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 5 Intubation at any time point after trial entry.
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 6 Duration of mechanical ventilation via an endotracheal tube (days).
| Study | HFNC (median, Interquartile range) | CPAP (median, IQR) |
| Kugelman 2015 | 4, 1.0‐15.0 | 2, 0.3‐6.5 |
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 7 Duration of any respiratory support (days).
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 8 Duration of hospitalisation (days).
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 9 Pneumothorax.
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 10 Nosocomial Sepsis.
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 11 Gastrointestinal perforation or severe NEC.
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 12 Days to full feeds.
Comparison 2 HFNC versus NIPPV for primary respiratory support after birth, Outcome 13 Nasal trauma.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 1 Death or CLD.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 2 CLD.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 3 Death.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 4 Treatment failure within 7 days of trial entry.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 5 Reintubation within 7 days of trial entry.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 6 Reintubation at any time point after trial entry.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 7 Duration of mechanical ventilation via an endotracheal tube (days).
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 8 Duration of any respiratory support (days after randomisation).
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 9 Duration of any respiratory support (postmenstrual age at end).
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 10 Duration of supplemental oxygen (days after randomisation).
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 11 Duration of supplemental oxygen (postmenstrual age at end).
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 12 Duration of hospitalisation (days).
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 13 Pneumothorax.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 14 Nosocomial Sepsis.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 15 ROP (any stage).
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 16 Gastrointestinal perforation or severe NEC.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 17 Days to full feeds.
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 18 Weight gain prior to discharge from hospital (grams).
Comparison 3 HFNC versus CPAP to prevent extubation failure, Outcome 19 Nasal trauma.
Comparison 4 Humidified HFNC versus non‐humidified HFNC to prevent extubation failure, Outcome 1 Reintubation within 7 days of trial entry.
Comparison 5 Alternative HFNC models to prevent extubation failure, Outcome 1 Death or CLD.
Comparison 5 Alternative HFNC models to prevent extubation failure, Outcome 2 CLD.
Comparison 5 Alternative HFNC models to prevent extubation failure, Outcome 3 Death.
Comparison 5 Alternative HFNC models to prevent extubation failure, Outcome 4 Treatment failure within 7 days of trial entry.
Comparison 5 Alternative HFNC models to prevent extubation failure, Outcome 5 Reintubation within 7 days of trial entry.
Comparison 5 Alternative HFNC models to prevent extubation failure, Outcome 6 Necrotising Enterocolitis.
Comparison 6 Alternative HFNC humidification devices, Outcome 1 Death or CLD.
Comparison 6 Alternative HFNC humidification devices, Outcome 2 CLD.
Comparison 6 Alternative HFNC humidification devices, Outcome 3 Treatment failure.
Comparison 6 Alternative HFNC humidification devices, Outcome 4 Intubation at any time point after trial entry.
Comparison 6 Alternative HFNC humidification devices, Outcome 5 Nasal trauma.
Comparison 7 HFNC for weaning from CPAP, Outcome 1 Death or CLD.
Comparison 7 HFNC for weaning from CPAP, Outcome 2 CLD.
Comparison 7 HFNC for weaning from CPAP, Outcome 3 Death.
Comparison 7 HFNC for weaning from CPAP, Outcome 4 Treatment failure.
Comparison 7 HFNC for weaning from CPAP, Outcome 5 Intubation at any time point after trial entry.
| Study | HFNC | CPAP |
| Abdel Hady 2011 | median 18 days (IQR 11.5‐29) | median 10.5 (IQR 4‐21) |
Comparison 7 HFNC for weaning from CPAP, Outcome 6 Duration of any respiratory support (days).
| Study | HFNC | CPAP |
| Abdel Hady 2011 | median (interquartile range): 14 (7.5–19.25) | median (interquartile range): 5 (1–8) |
| Badiee 2015 | mean 20.6 +/‐16.8 hours | mean 49.5 +/‐ 25.3 hours |
Comparison 7 HFNC for weaning from CPAP, Outcome 7 Duration of oxygen supplementation (days).
Comparison 7 HFNC for weaning from CPAP, Outcome 8 Duration of hospitalisation (days).
Comparison 7 HFNC for weaning from CPAP, Outcome 9 Pneumothorax.
Comparison 7 HFNC for weaning from CPAP, Outcome 10 Necrotising enterocolitis.
Comparison 7 HFNC for weaning from CPAP, Outcome 11 Nosocomial sepsis.
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Death or CLD Show forest plot | 4 | 439 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.66 [0.56, 4.94] |
| 1.1 < 28 weeks | 0 | 0 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 1.2 28 ‐ 32 weeks | 1 | 37 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.55 [0.29, 22.31] |
| 1.3 ≥ 32 weeks | 1 | 88 | Risk Ratio (M‐H, Fixed, 95% CI) | 6.54 [0.32, 132.33] |
| 1.4 < 37 weeks' (subgroup data not available) | 3 | 314 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.76 [0.15, 3.77] |
| 2 CLD Show forest plot | 4 | 439 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.07 [0.64, 6.64] |
| 2.1 < 28 weeks | 0 | 0 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 2.2 28 ‐ 32 weeks | 1 | 37 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.55 [0.29, 22.31] |
| 2.3 > 32 weeks | 1 | 88 | Risk Ratio (M‐H, Fixed, 95% CI) | 6.54 [0.32, 132.33] |
| 2.4 < 37 weeks' (subgroup data not available) | 3 | 314 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.06 [0.19, 5.97] |
| 3 Death Show forest plot | 4 | 439 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.36 [0.01, 8.73] |
| 3.1 < 28 weeks | 0 | 0 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 3.2 28 ‐ 32 weeks | 1 | 37 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 3.3 > 32 weeks | 1 | 88 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 3.4 < 37 weeks' (subgroup data not available) | 3 | 314 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.36 [0.01, 8.73] |
| 4 Treatment failure within 7 days of trial entry Show forest plot | 4 | 439 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.30 [0.73, 2.34] |
| 4.1 < 28 weeks | 0 | 0 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 4.2 28 ‐ 32 weeks | 1 | 37 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.17 [0.01, 3.34] |
| 4.3 > 32 weeks | 1 | 88 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.13 [0.41, 3.08] |
| 4.4 < 37 weeks' (subgroup data not available) | 3 | 314 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.77 [0.81, 3.89] |
| 5 Intubation within 7 days of trial entry Show forest plot | 2 | 247 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.38 [0.86, 6.57] |
| 6 Intubation at any time point after trial entry Show forest plot | 1 | 125 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.16 [0.49, 2.71] |
| 7 Duration of any respiratory support (days) Show forest plot | 2 | 192 | Mean Difference (IV, Fixed, 95% CI) | 0.97 [0.08, 1.86] |
| 8 Duration of supplemental oxygen (days) Show forest plot | 1 | 125 | Mean Difference (IV, Fixed, 95% CI) | 3.70 [‐0.66, 8.06] |
| 9 Duration of hospitalisation (days) Show forest plot | 3 | 262 | Mean Difference (IV, Fixed, 95% CI) | 0.50 [‐0.55, 1.56] |
| 10 Pneumothorax Show forest plot | 3 | 369 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.54 [0.17, 1.79] |
| 11 Nosocomial sepsis Show forest plot | 4 | 439 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.29 [0.66, 2.54] |
| 12 Gastrointestinal perforation or severe NEC Show forest plot | 1 | 125 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.46 [0.14, 83.27] |
| 13 Days to full feeds Show forest plot | 2 | 195 | Mean Difference (IV, Fixed, 95% CI) | ‐0.41 [‐1.46, 0.63] |
| 14 Nasal trauma Show forest plot | 3 | 258 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.62 [0.34, 1.15] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Death or CLD Show forest plot | 1 | 75 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.13 [0.30, 4.32] |
| 1.1 < 28 weeks' | 1 | 3 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.5 [0.38, 6.00] |
| 1.2 28 ‐ 32 weeks' | 1 | 28 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.75 [0.05, 10.82] |
| 1.3 ≥ 32 weeks' | 1 | 44 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 2 CLD Show forest plot | 1 | 75 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.13 [0.30, 4.32] |
| 2.1 < 28 weeks' | 1 | 3 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.5 [0.38, 6.00] |
| 2.2 28 ‐ 32 weeks' | 1 | 28 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.75 [0.05, 10.82] |
| 2.3 ≥ 32 weeks' | 1 | 44 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 3 Death Show forest plot | 1 | 75 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 3.1 < 28 weeks' | 1 | 3 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 3.2 28 ‐ 32 weeks' | 1 | 28 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 3.3 ≥ 32 weeks' | 1 | 44 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 4 Treatment failure within 7 days of trial entry Show forest plot | 1 | 75 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.92 [0.50, 1.69] |
| 4.1 < 28 weeks' | 1 | 3 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.0 [0.39, 2.58] |
| 4.2 28 ‐ 32 weeks' | 1 | 28 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.9 [0.36, 2.26] |
| 4.3 ≥ 32 weeks' | 1 | 44 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.91 [0.33, 2.55] |
| 5 Intubation at any time point after trial entry Show forest plot | 1 | 76 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.85 [0.44, 1.65] |
| 6 Duration of mechanical ventilation via an endotracheal tube (days) Show forest plot | 1 | 76 | Mean Difference (IV, Fixed, 95% CI) | 0.5 [‐1.64, 2.64] |
| 7 Duration of any respiratory support (days) Show forest plot | Other data | No numeric data | ||
| 8 Duration of hospitalisation (days) Show forest plot | 1 | 76 | Mean Difference (IV, Fixed, 95% CI) | 1.0 [‐9.10, 11.10] |
| 9 Pneumothorax Show forest plot | 1 | 76 | Risk Ratio (M‐H, Fixed, 95% CI) | 5.0 [0.25, 100.80] |
| 10 Nosocomial Sepsis Show forest plot | 1 | 76 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.33 [0.32, 5.56] |
| 11 Gastrointestinal perforation or severe NEC Show forest plot | 1 | 76 | Risk Ratio (M‐H, Fixed, 95% CI) | 5.0 [0.25, 100.80] |
| 12 Days to full feeds Show forest plot | 1 | 76 | Mean Difference (IV, Fixed, 95% CI) | 1.70 [‐1.21, 4.61] |
| 13 Nasal trauma Show forest plot | 1 | 76 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Death or CLD Show forest plot | 5 | 896 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.94 [0.78, 1.14] |
| 1.1 < 28 weeks | 2 | 233 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.01 [0.79, 1.29] |
| 1.2 28 ‐ 32 weeks | 5 | 384 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.78 [0.57, 1.08] |
| 1.3 ≥ 32 weeks | 3 | 279 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.29 [0.67, 2.48] |
| 2 CLD Show forest plot | 5 | 893 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.96 [0.78, 1.18] |
| 2.1 < 28 weeks | 2 | 233 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.04 [0.80, 1.36] |
| 2.2 28 ‐ 32 weeks | 5 | 382 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.82 [0.57, 1.17] |
| 2.3 ≥ 32 weeks | 3 | 278 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.18 [0.52, 2.70] |
| 3 Death Show forest plot | 5 | 896 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.77 [0.43, 1.36] |
| 3.1 < 28 weeks | 2 | 233 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.75 [0.25, 2.29] |
| 3.2 28 ‐ 32 weeks | 5 | 384 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.54 [0.22, 1.31] |
| 3.3 ≥ 32 weeks | 3 | 279 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.23 [0.43, 3.48] |
| 4 Treatment failure within 7 days of trial entry Show forest plot | 5 | 786 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.21 [0.95, 1.55] |
| 4.1 < 28 weeks | 2 | 233 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.22 [0.91, 1.64] |
| 4.2 28 ‐ 32 weeks | 4 | 342 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.80 [0.44, 1.44] |
| 4.3 ≥ 32 weeks | 2 | 171 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.25 [0.56, 2.79] |
| 4.4 < 37 weeks' (subgroup data not available) | 1 | 40 | Risk Ratio (M‐H, Fixed, 95% CI) | 4.0 [1.33, 12.05] |
| 5 Reintubation within 7 days of trial entry Show forest plot | 6 | 934 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.91 [0.68, 1.20] |
| 5.1 < 28 weeks | 2 | 233 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.85 [0.57, 1.26] |
| 5.2 28 ‐ 32 weeks | 5 | 382 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.51 [0.27, 0.97] |
| 5.3 ≥ 32 weeks | 3 | 279 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.05 [0.56, 1.97] |
| 5.4 < 37 weeks (subgroup data not available) | 1 | 40 | Risk Ratio (M‐H, Fixed, 95% CI) | 4.0 [1.33, 12.05] |
| 6 Reintubation at any time point after trial entry Show forest plot | 4 | 746 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.86 [0.67, 1.10] |
| 7 Duration of mechanical ventilation via an endotracheal tube (days) Show forest plot | 1 | 303 | Mean Difference (IV, Fixed, 95% CI) | ‐3.70 [‐6.74, ‐0.66] |
| 8 Duration of any respiratory support (days after randomisation) Show forest plot | 2 | 529 | Mean Difference (IV, Fixed, 95% CI) | ‐0.70 [‐4.14, 2.74] |
| 9 Duration of any respiratory support (postmenstrual age at end) Show forest plot | 2 | 424 | Mean Difference (IV, Fixed, 95% CI) | ‐0.39 [‐1.09, 0.32] |
| 10 Duration of supplemental oxygen (days after randomisation) Show forest plot | 2 | 519 | Mean Difference (IV, Fixed, 95% CI) | 1.54 [‐3.42, 6.51] |
| 11 Duration of supplemental oxygen (postmenstrual age at end) Show forest plot | 2 | 433 | Mean Difference (IV, Fixed, 95% CI) | ‐0.97 [‐1.87, ‐0.07] |
| 12 Duration of hospitalisation (days) Show forest plot | 2 | 518 | Mean Difference (IV, Fixed, 95% CI) | 0.90 [‐4.17, 5.98] |
| 13 Pneumothorax Show forest plot | 5 | 896 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.35 [0.11, 1.06] |
| 14 Nosocomial Sepsis Show forest plot | 2 | 529 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.92 [0.59, 1.43] |
| 15 ROP (any stage) Show forest plot | 2 | 343 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.90 [0.40, 2.07] |
| 16 Gastrointestinal perforation or severe NEC Show forest plot | 5 | 840 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.52 [0.24, 1.11] |
| 17 Days to full feeds Show forest plot | 3 | 387 | Mean Difference (IV, Fixed, 95% CI) | 0.60 [0.35, 0.85] |
| 18 Weight gain prior to discharge from hospital (grams) Show forest plot | 2 | 518 | Mean Difference (IV, Fixed, 95% CI) | 66.32 [‐45.63, 178.27] |
| 19 Nasal trauma Show forest plot | 4 | 645 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.64 [0.51, 0.79] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Reintubation within 7 days of trial entry Show forest plot | 1 | 28 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.18 [0.01, 3.34] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Death or CLD Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 2 CLD Show forest plot | 1 | 39 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.86 [0.29, 2.58] |
| 3 Death Show forest plot | 1 | 40 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.44 [0.02, 10.29] |
| 4 Treatment failure within 7 days of trial entry Show forest plot | 1 | 40 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.35 [0.31, 5.90] |
| 5 Reintubation within 7 days of trial entry Show forest plot | 1 | 40 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.97 [0.37, 2.53] |
| 6 Necrotising Enterocolitis Show forest plot | 1 | 40 | Risk Ratio (M‐H, Fixed, 95% CI) | 4.0 [0.17, 92.57] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Death or CLD Show forest plot | 1 | 60 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.5 [0.53, 11.89] |
| 2 CLD Show forest plot | 1 | 60 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.5 [0.53, 11.89] |
| 3 Treatment failure Show forest plot | 1 | 60 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.0 [0.33, 27.23] |
| 4 Intubation at any time point after trial entry Show forest plot | 1 | 60 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.0 [0.07, 15.26] |
| 5 Nasal trauma Show forest plot | 1 | 60 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.5 [0.27, 8.34] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Death or CLD Show forest plot | 2 | 148 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.0 [0.49, 18.50] |
| 2 CLD Show forest plot | 2 | 148 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.05 [0.50, 18.73] |
| 2.1 28 ‐ 32 weeks | 1 | 37 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.16 [0.14, 72.84] |
| 2.2 ≥ 32 weeks | 1 | 23 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 2.3 < 37 weeks (subgroup data not available) | 1 | 88 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.0 [0.32, 27.74] |
| 3 Death Show forest plot | 2 | 148 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 4 Treatment failure Show forest plot | 2 | 148 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.3 [0.59, 2.88] |
| 4.1 28 ‐ 32 weeks | 1 | 37 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.23 [0.51, 2.97] |
| 4.2 ≥ 32 weeks | 1 | 23 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 4.3 < 37 weeks (subgroup data not available) | 1 | 88 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.5 [0.26, 8.54] |
| 5 Intubation at any time point after trial entry Show forest plot | 2 | 148 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.11 [0.01, 2.00] |
| 6 Duration of any respiratory support (days) Show forest plot | Other data | No numeric data | ||
| 7 Duration of oxygen supplementation (days) Show forest plot | Other data | No numeric data | ||
| 8 Duration of hospitalisation (days) Show forest plot | 2 | 148 | Mean Difference (IV, Fixed, 95% CI) | ‐3.31 [‐6.62, 0.00] |
| 9 Pneumothorax Show forest plot | 1 | 60 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 10 Necrotising enterocolitis Show forest plot | 2 | 148 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.33 [0.01, 7.97] |
| 11 Nosocomial sepsis Show forest plot | 2 | 148 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.88 [0.55, 1.39] |
